PONDERACIÓN DE LAS VARIABLES DEL CLIMA ORGANIZACIONAL

In addition to using an isolation requirement on the 2017 trigger, we also use isolation in the offline selection of events. Isolation ensures that there is not too much activity in the detector around the photon from both other interaction vertices but also from other decay product originating from the primary vertex. Non-isolated candidates have a higher probability of being misconstructed as photons. Two types of isolation variables are used; track-based and calorimeter-based. Track-based isolation, ptcone20, is defined as the scalar sum of the trans- verse momenta of all tracks withpT >1GeV in the cone of size ∆R=

p

((∆η)2_+(∆φ)2_{) = 0.2}

around each photon candidate. Only tracks that are reconstructed as originating from the primary vertex are used, and tracks associated with conversions are also removed. The calorimeter-based isolation uses only the topological clusters in the calorimeter. topoetcone40 is the sum of all positive energy in clusters within a cone of ∆R= 0.4 after having subtracted the contribution from the photon candidate itself.

Similar to how the photon ID is done, there are three different isolation cuts that are available as working points:

• FixedCutTightCaloOnly: topoetcone40<0.022pT+ 2.45 [GeV], aiming for SM measure-

ments;

• FixedCutTight: topoetcone40<0.022pT+ 2.45 [GeV] andptcone20/pT<0.05, aiming

for high-ETphotons;

• FixedCutLoose: topoetcone20<0.065pTandptcone20/pT<0.05, designed for the SM H_→γγ analysis.

In order to choose which working point to use, we look at the significance for each cut. We will measure the significance relative to the FixedCutLoose working point by taking the ratio

Generator of signal samples Mass point [GeV]

MadGraph MX= 40,50,60,70,80,90,100,110 MX = 120,140,160,180,200

MX−5< mγγ < MX+ 5 GeV MX−10< mγγ< MX+ 10 GeV

PowHeg+Pythia8 MX = 40,60,80,100,120 MX = 160

MX−10< mγγ < MX+ 10 GeV MX−20< mγγ< MX+ 20 GeV

Table 5.5: mγγ cuts applied on signal and background samples with respect to the different

mass of resonance points.

of the two. This gives us the relative efficiency relative = Ntestcut/NF ixedCutLoose where

Ntestcut (NF ixedCutLoose) is the number of events passing the cut we are testing (number

of events passing FixedCutLoose). The gain or loss in significance as measured relative to FixedCutLoose is written as ZtestCut/ZF ixedCutLoose = S/√B where S and B are the

relative efficiencies for the signal and background respectively.

The samples primarily used for the isolation study are the MadGraph samples, however, samples produced with PowHeg+Pythia8 are used as well to check the effect of the working points on different production modes. These samples are detailed in section 5.2.2. For the background samples we use the Sherpa leading order samples in addition to the Sherpaγ jet samples. Theγjet component ranges from about 50% to 30% of the total background events according to the study in section 5.6.1. This means we need to account for this contribution as jets have a different isolation distribution from photons. Isolation working points are tested for two fraction of γ jet contribution, 50% γγ : 50% γ jet and 70% γγ: 30% γ jet. Jet-jet components are neglegted as they are small (_∼6% section 5.6.1).

The events that are used for testing are required to pass the basic kinematic selection of 2g20 tight trigger, tight identification and ET > 22 GeV. There is then an addition cut

placed on the events invariant mass in a window around the mass of the signal being tested. These cuts range from_±5GeV for lower masses to_±20 GeV for higher masses in the PowHeg samples. TheseMγγ cuts are shown in table 5.5.

Figure 5.3 shows the relative isolation efficiencies for the background samples. The binning in this plot corresponds to the mass points in the MadGraph signal sample so that they may be easily compared and combined. Likewise, figure 5.4 shows the same background samples, but with the binning corresponding to the mass points in the PowHeg signal samples. As can be seen, the ratios for the FixedCutTight isolation are below 1, showing that this working point has a higher rejection than the FixedCutLoose. This is expected as the tight cuts are designed to have a greater rejection (or lower efficiency). The relative efficiency for the FixedCutTightCaloOnly is greater than or close to 1, showing that its rejection power is lower than that of the track-based FixedCutLoose working point.

The relative efficiency for the MadGraph signal samples are shown in firgure 5.5. We find a similar situation to the background efficiency with the FixedCutTightCaloOnly being more efficient than FixedCutLoose, and FixedCutTight being slightly less. Figure 5.6 shows the significance relative to the FixedCutLoose working point for the two different fractions of γ-jet. In both cases, the FixedCutTight working point has a higher significance than the calorimeter only version, but both are lower and 1 in general. Although the FixedCutTight isolation is very close to one for a few mass points, it is on average lower.

Figure 5.7 shows the relative efficiencies for the PoweHeg + Pythia8 signal samples with the ggH, ttH, VBF, WH, and ZH production modes. The significance for each of these production modes is shown in figure 5.8 for the 70% : 30%γ-jet fraction, and the same in figure 5.9 for the 50% : 50%γ-jet fraction. We see similar results for the PowHeg + Pythia8 samples as for the MadGraph samples. The FixedCutLoose working point tends to have the highest significance across the invariant mass range and so it is used for the analysis.

(a) (b)

(c) (d)

Figure 5.3: Isolation efficiency relative to the FixedCutLoose working point for background samples: (a) γγ. (b) γjet. (c) combined background sample with γγ : γjet = 70% : 30%. (d) γγ : γjet = 50% : 50%. Efficiencies for the FixedCutTight (respectively FixedCut- TightCaloOnly) working point are represented with full (respectively open) circles.

(a) (b)

(c) (d)

Figure 5.4: Isolation efficiency relative to the FixedCutLoose working point for background samples: (a) γγ. (b) γjet. (c) combined background sample with γγ : γjet = 70% : 30%. (d) γγ : γjet = 50% : 50%. Efficiencies for the FixedCutTight (respectively FixedCut- TightCaloOnly) working point are represented with full (respectively open) circles.

(a)

Figure 5.5: Isolation efficiency relative to the FixedCutLoose working point for MadGraph signal samples. Black open circles represents that FixedCutTight isolation requirement is applied, and blue dots represents FixedCutTightCaloOnly.

(a) (b)

Figure 5.6: Significance relative to the FixedCutLoose working point with MadGraph signal sample and background contains: (a) γγ : γjet = 70% : 30%. (b)γγ : γjet = 50% : 50%. Significances for the FixedCutTight (respectively FixedCutTightCaloOnly) working point are represented with full (respectively open) circles.

(a) (b)

(c) (d)

(e)

Figure 5.7: Isolation efficiency relative to the FixedCutLoose working point for signal samples: (a) ggH. (b) ttH. (c) VBFH. (d) WH. (e) ZH. Efficiencies for the FixedCutTight (respectively FixedCutTightCaloOnly) working point are represented with full (respectively open) circles.

(a) (b)

(c) (d)

(e)

Figure 5.8: Significance relative to the FixedCutLoose working point with signal sample as: (a) ggH. (b) ttH. (c) VBFH. (d) WH. (e) ZH. The background containsγγ:γjet= 70% : 30%. Significances for the FixedCutTight (respectively FixedCutTightCaloOnly) working point are represented with full (respectively open) circles.

(a) (b)

(c) (d)

(e)

Figure 5.9: Significance relative to the FixedCutLoose working point with signal sample as: (a) ggH. (b) ttH. (c) VBFH. (d) WH. (e) ZH. The background containsγγ:γjet= 50% : 50%. Significances for the FixedCutTight (respectively FixedCutTightCaloOnly) working point are represented with full (respectively open) circles.